How the analysis and visualizations are made

Every map, chart, rendering, and analysis on this site starts from a source record, and each can be traced back to it. This page sets out how the work is made, what it can and cannot show, and how to report an error.

Analysis

The analysis follows four steps:

1. Document compilation. Assemble the relevant site plans, spreadsheets, public records, agency presentations, and historical files.
2. Technical review. Check capacity, cost, and service assumptions against agency rules, planning constraints, and comparable systems.
3. Spatial visualization. Build maps, diagrams, and 3D models directly from the documented spatial constraints.
4. Presentation. Turn the findings and alternatives into briefs, decks, or public pages.

Operating claims come first. Train counts, dwell assumptions, turnback constraints, platform loads, service patterns, and station geometry decide whether a plan can function, so each is checked against the source before anything is drawn.

Data tools

The interactive tools use public data: Census journey-to-work and American Community Survey (ACS) tables, Longitudinal Employer-Household Dynamics (LEHD) commuting data, the National Transit Database, agency spending records, and public geographic files. Estimates drawn from survey data carry margins of error. The congestion pricing explorer, for example, reports them on each two-page district fact sheet. As each tool page is rebuilt on the site’s tool template, it names its data source, coverage, and update date and lists its limits.

Visualizations and renderings

Every rendering starts from public plans, aerial references, and site constraints. A visual alternative begins with an explicit policy claim, a defined geography, a plausible path to implementation, and a clear line between known facts and assumptions.

AI tools speed up production: they generate variations and framings, and the chosen images are then composited and refined by hand. Each output is checked against the source plans before it is published. AI coding agents have also been used to build interactive models, such as the BQE corridor 3D model.

Every rendering is published with a disclosure. The images show proposals, not engineering designs. They do not establish structural feasibility, traffic performance, cost, or phasing, and they are not architectural services, engineering drawings, construction documents, substitutes for environmental review, or official agency plans.

Records

Public records come from requests to city, state, and federal agencies, including New York’s Freedom of Information Law (FOIL), and from published agency documents. Where a page quotes a record, the quotation is verbatim and linked to its source, and where a page characterizes an organization’s conduct, it cites the document the characterization rests on.

Corrections

To report an error, email liam@liamblank.com with the page and the problem. Confirmed errors are corrected on the page, and the correction is noted there with its date.